Our client is a prominent player in the renewable energy sector, with a specific focus on wind power. They manage an extensive array of wind turbines across various regions, supplying local citizens and manufacturing facilities with electricity. Detailed information about the client cannot be disclosed under the provisions of the NDA.

Challenge

Power outages and costly repairs for customer's wind farm

The renewable energy space, especially wind power, is dynamic and requires permanent innovation to ensure maximum efficiency and uptime. Having operated in this field for over 20 years, our client has encountered many unexpected breakdowns, resulting in power outages and costly repairs. With ambitious expansion plans, they sought IoT energy solutions to monitor real-time wind turbine performance and prevent malfunctions through intelligent ML algorithms. The customer charged Innowise with the task of developing an IoT for energy management solution that could offer real-time monitoring and predictive analysis to ensure their wind turbines operate efficiently and safely around the clock.

Solution

IoT energy management solution that predicts energy production and prevents errors

Based on the customer’s requirements and expectations, Innowise came up with an IoT & ML-driven solution that predicts energy production based on the information accumulated from meteorological sensors and turbines. Our project team developed an advanced platform that offers real-time information about each wind turbine’s status, facilitating instant decision-making and responding to operational cues without delay.

Programmable logic controllers (PLC)

As the cornerstone of automation, we used PLCs to collect data from sensors installed throughout the wind turbines. These sensors measure a wide range of operational parameters such as wind speed, turbine rotational speed, temperature, vibration levels, and torque. By processing this data, PLCs give an accurate, real-time picture of the wind turbine’s performance, detect malfunctions, and analyze energy production efficiency.

Sensor indicators deviating from predefined thresholds — like an unexpected temperature increase or vibration level — signal potential issues such as mechanical wear, lubrication needs, or component failure. PLCs, in turn, recognize these patterns and trigger alarms or shut down the turbine to prevent damage. Furthermore, PLCs record the power output data and analyze it along with wind conditions to determine if the turbines are generating power efficiently. Then they flag an anomaly if the wind speed is optimal but energy output is below the threshold, indicating an issue like blade deterioration, misalignment, etc. Through PLS-enabled timely maintenance and malfunction prevention, well-balanced energy production ensures the longevity of the equipment.

Data lake

Since our client has dozens of wind turbines scattered across distinct regions, our developers were tasked to build a robust data lake to store massive event-driven messages. We created a central repository where data from all the turbines, regardless of geographical location, is collected and stored. This includes not just structured data but also unstructured and semi-structured data, like logs, sensor readings, images, and more. IoT specialists ensured that all data nuances were preserved, allowing for more detailed analysis and reducing data loss risks.

Also, our project team enabled concurrent data processing across multiple nodes. This means that large datasets can be processed in parallel, significantly speeding up analysis and reporting tasks. This is crucial for predictive maintenance, where time-sensitive insights can prevent costly downtime and sudden breakdowns of the wind turbines. The data for analysis is retrieved from the PLCs, then stored and processed by AWS IoT Core and Lambda functions.

Data vizualization

To visualize data, our project team opted for vivid Grafana dashboards. We set up dashboards comprised of various visual elements tailored to the needs of IoT energy management. As a result, operational managers, for example, can have an overview of real-time turbine performance, while maintenance teams can take a more detailed look at wear-and-tear indicators with Grafana.Thus, linear charts show trends over time, like power output throughout the day. Map charts provide geographical visualizations of the turbines’ locations, enabling a quick overview of the entire wind farm’s status. Time series predict future trends based on past and present data, essential for planning and forecasting. Histograms detail the distribution of specific variables, such as wind speeds or turbine output, which is helpful for statistical analysis. Finally, geomaps layer additional data on geographical maps, such as weather patterns, to measure the influence of unfavorable weather conditions.Overall, the client gets a transparent and informative visualization of IoT data that is easily interpreted and acted upon. For example, through color-coded indicators, a maintenance technician can easily spot a turbine operating outside of its optimal range and take proactive measures to eliminate malfunction.

Analytical reports

Further, our engineers ensured the IoT-driven platform generates analytical reports to deliver comprehensive insights about wind turbines’ performance. This data helps identify which turbines operate well and which may require maintenance or adjustments. Besides that, the IoT-based system uses historical and real-time data for predictive maintenance to forecast future outcomes under different conditions. In this way, it recommends when to schedule maintenance or optimize operations without waiting for an issue to occur.

Additionally, by analyzing performance trends and external factors like weather conditions, the system proposes scenarios where IoT energy management can be optimized. For instance, it suggests ways to optimize energy consumption, reduce extra expenses, determine the ideal times for harvesting wind energy, manage storage effectively, sell excess energy back to the grid, and streamline maintenance procedures.

Error prediction

Using the power of data science (DS) and machine learning operations (MLOps), we developed a predictive model that analyzes various factors affecting turbine health, such as vibration levels, temperature, and performance metrics. This model continually learns from incoming data, enabling it to identify patterns that precede equipment failures. When it detects these early warning signs, it triggers an alert system, allowing maintenance teams to address issues proactively before they lead to breakdowns.

Technologies & tools

Front-end

JavaScript, React, Redux

Back-end

Python, FastAPI

DE/ML

Apache Spark

Cloud

AWS EKS, AWS ECS, AWS ECR, AWS EC2, AWS API Gateway, AWS IOT Core, AWS Kinesis, AWS Lake Formation, AWS Lambda, AWS RDS Postgres, AWS TimeStream DB; AWS S3, AWS Route 53; AWS CloudFront

DevOps

Kubernetes, Docker, AWS EKS, AWS ECS

Database

PostgreSQL, AWS TimeStream

Visualization

Grafana

Process

Developing a custom IoT-based system for monitoring and maintaining wind turbines was a complex yet rewarding journey. We began with extensive discussions with our client to understand their needs and challenges. This phase involved identifying the core functionalities required for the IoT system, such as real-time monitoring, error prediction, and data analytics.With the requirements in hand, we developed a comprehensive project plan outlining the timeline, resources, budget, and risk management strategies. Our development phase involved creating the system architecture and the user interface, including custom algorithms for data analysis, visualizations, predictive maintenance, and integrated PLCs and AWS IoT Core. Agile methodology enabled us to adapt quickly and effectively to changing requirements and feedback throughout the project. Regular stand-ups, sprint reviews, and retrospectives were integral to our process, fostering a collaborative and dynamic work environment. This approach enabled us to deliver a tailored, robust, efficient IoT-based system aligned perfectly with our client’s unique needs.As of now, Innowise provides post-lunch maintenance and support, fixes minor bugs, and releases regular updates.

Team

Project Manager

Business Analyst

Solution Architect

Front-End Developer

Back-End Developers

Embedded Developer

ML Developer

DE Developer

DevOps

QA Engineers

Stakeholder’s SME

Results

18% reduction in maintenance and repair costs with IoT & ML-driven system

Innowise has built an IoT & ML-driven scalable system that predicts energy production based on the system of programmable logic controllers. We developed a sophisticated platform that gathers critical information from the wind turbines, assesses their performance and provides accurate insights for informed decision-making. Based on this information, customer managers can monitor turbines’ conditions in real-time and suggest scenarios to optimize energy production and reduce superfluous expenses. Due to ML algorithms, our ground-breaking solution predicts power generation based on weather forecasts and accumulated analytics. Furthermore, it determines the best time to shut down wind farms and conduct maintenance accordingly. This is particularly crucial for turbines in remote or harsh environments where repairs can be challenging and expensive.

Project duration

September 2021 - Ongoing

up to 6%

increase in energy production

18%

reduction in maintenance and repair costs

critical threats prevented

Related cases

Contact us!

Book a call or fill out the form below and we’ll get back to you once we’ve processed your request.

What happens next?

Having received and processed your request, we will get back to you shortly to detail your project needs and sign an NDA to ensure the confidentiality of information.

After examining requirements, our analysts and developers devise a project proposal with the scope of works, team size, time, and cost estimates.

We arrange a meeting with you to discuss the offer and come to an agreement.

We sign a contract and start working on your project as quickly as possible.

IoT for energy management: up to 6% increase in energy production

Customer

Challenge

Power outages and costly repairs for customer's wind farm

Solution